Does Growth Hormone Cause Cancer?

P. J. Jenkins; A. Mukherjee; S. M. Shalet


Clin Endocrinol. 2006;64(2):115-121. 

In This Article

Cancer Risk in Children Treated With GH

Is There an Increased Risk of Tumour Recurrence?

A specific group of patients at risk of GHD include the long-term survivors of childhood cancers who have received treatment with cranial irradiation either for primary brain tumours, nasopharyngeal carcinoma or acute lymphoblastic leukaemia, or total body irradiation in preparation for bone marrow transplantation.[19] Although newer treatment strategies have decreased mortality rates substantially such that the overall 5-year survival rate for childhood cancer is in excess of 70%, survivors are at risk of developing a variety of late complications that are directly attributable to their previous cancer treatment.

Brain tumour recurrence is a frequent cause of death in patients treated with GH,[20] but there is now a moderate amount of information on whether recurrence rates are greater after this treatment than in comparable untreated patients. A large study by Swerdlow and colleagues followed 180 children with brain tumours attending three large hospitals in the UK and treated with GH during 1965-1996, and 891 children with brain tumours who received radiotherapy but not GH in these hospitals.[21] Thirty-five first recurrences occurred in the GH-treated children and 434 in the untreated children. The relative risk of first recurrence in those treated with GH compared with untreated patients, adjusted for potentially confounding prognostic variables, was decreased (0·6; 95% confidence interval (CI), 0·4–0·9), as was the relative risk of mortality (0·5; 95% CI, 0·3–0·8). There was no significant trend in relative risk of recurrence with cumulative time over which GH treatment had been given or with time elapsed since this treatment started. The relative risk of mortality increased significantly with time from the first GH treatment. The results, based on much larger numbers than earlier studies on this subject, suggest that GH does not increase the risk of recurrence of childhood brain tumours. Data from the Childhood Cancer Survivor Study (CCSS) are also consistent with the finding that GH does not increase risk of disease recurrence for both primary brain tumours and acute leukaemias.[22]

However, despite these reassuring data, the rising trend in mortality relative risks with longer follow-up in the Swerdlow study indicates the need for continued surveillance. It is not possible to state with 100% confidence that GH therapy after cancer treatment is absolutely safe and so the need for ongoing surveillance programs continues.

Second Malignant Neoplasms

A retrospective cohort of 13 581 children, diagnosed with common cancers in 25 USA/Canadian institutes, before the age of 21 years and who had survived at least 5 years demonstrated a statistically significant excess of second malignant neoplasms (SMNs) following all childhood cancers.[23] The effect of GH treatment on incidence of SMNs in such individuals is therefore important to determine as they are a group already at increased risk of malignancy. In 2002, Sklar et al., using a time-dependent Cox model, compared the risk of disease recurrence, SMNs and death in 361 cancer survivors treated with GH (including 172 brain tumour survivors) from among 13 539 survivors enrolled in the CCSS, with survivors who were not treated with GH.[22] The relative risk of disease recurrence was found to be 0·83 (95% CI, 0·37-1·86; P = 0·65) for survivors treated with GH. The relative risk of recurrence was not increased for any of the major cancer diagnoses. Subjects treated with GH were diagnosed with 15 SMNs, all solid tumours and no secondary leukaemias, for an overall relative risk of 3·21 (95% CI, 1·88-5·46; P < 0·0001). This was mainly caused by a small excess number of SMN observed in survivors of acute leukaemia treated with GH: osteogenic sarcoma in three of the leukaemias/lymphoma survivors treated with GH (n = 122) vs only two cases in more than 4500 non-GH-treated leukaemia/ lymphoma survivors. The risk of death was not associated with GH use (P = 0·43). The authors concluded that GH therapy does not appear to increase the risk of disease recurrence or death in survivors of childhood cancer. The increased number of SMN, particularly in survivors of acute leukaemia, is of concern, but the data need to be interpreted with caution, given the small number of events.

Is There an Increased Risk of De Novo Cancer With GH Therapy?

An initial early report of de novo leukaemia occurring in a child with possible Fanconi's anaemia after GH therapy[24] has not been substantiated by extensive long-term follow-up data, relating to GH replacement, from single centres and large multinational databases that have generally found no increase in the overall occurrence of de novo neoplasia or the rate of regrowth of primary pituitary tumours.[25,26,27]

In a study by Swerdlow et al., 1848 patients treated in childhood/ early adulthood from 1959 to 1985 with human pituitary GH were followed through for cancer incidence to December 1995 and for mortality through to December 2000[28] Risk of cancer was compared with that in the general population after controlling for age, sex and calendar period. The overall risk of cancer mortality was increased approximately threefold and from colorectal cancer and Hodgkin's disease (HD) approximately 11-fold. The incidence of colorectal cancer was increased approximately eightfold and both incidences and mortality of colorectal cancer, as well as that of HD, were increased even after excluding patients whose original diagnoses gave them higher risk of cancer. However, only two deaths each of colorectal cancer and HD were reported. The relatively small cohort and small number of cases argue for caution in the rush to judgement or conclusion, but again indicate the need for on-going long-term large surveillance programs.

The existing data available in paediatrics do not support the notion of an excess of malignancy after GH treatment. In the two largest international databases and surveillance studies, with a total of some 86 000 patients on GH, representing almost 250 000 GH treatment years, there is only one report of a gastrointestinal carcinoma – an adenocarcinoma in a 15-year-old girl initially treated by radiotherapy and chemotherapy for a brain tumour (astrocytoma). She developed gastrointestinal bleeding 3·5 years after the start of GH therapy when the tumour was diagnosed. There is also a report of spontaneous colon cancer in a girl with Turner's syndrome many years after discontinuation of GH therapy.[29] Similar reassuring data are available in adults with the largest database, showing no significant increase in cancer incidence in the approximately 8000 patients treated for a total of 27 000 patient years.[27] However, the average duration of follow-up on GH treatment is only 4 years and longer-term surveillance is clearly needed. Furthermore, none of the surveillance groups has an adequate control group of untreated patients. However, given that GH is administered, certainly in adults, as a replacement therapy, with the aim being restoration of GH and IGF-1 levels to within age-matched normal ranges, clarification is needed as to which reference population cancer risk is being compared. One can imagine that long-term surveillance studies might indicate cancer incidence to be increased in patients with GHD treated with GH, but that this is likely to be a similar risk level as the general population and thus overall cancer risk will not be increased. Any risk analysis needs to take into account the overall benefits of GH compared to the proven morbidity associated with untreated GHD. In contrast with the use of GH for replacement purposes, there is much more concern regarding the unlicensed usage of rGH in both the elderly and normal adults for its anabolic effects and as an 'elixir of youth'. In these subjects, IGF-1 levels are often elevated above the normal range with consequent distortion of normal physiology.

Epidemiological Studies in the General Population

Considerable epidemiological data have suggested a possible link between circulating GH and/or IGF-1 levels and the development of a variety of different cancers. More than 20 years ago, Emerman et al. observed that women with breast cancer had elevated serum GH levels[30] and subsequently it was noted that patients with breast and prostate cancer had increased circulating IGF-1 levels.[31] However, these studies were limited by the potential influence of the cancer itself on the GH/IGF-1 axis or by the possible secretion of IGF-1 by the cancer. The more recent results from a number of prospective studies would appear to overcome these limitations.[10,32,33,34,35] Serum had been stored from large cohorts of subjects and IGF-1 and IGFBP-3 levels measured many years later in those subjects who had developed a cancer, and comparison made with a control group of subjects. Many of these studies have suggested that subjects with serum IGF-1 levels that are in the higher centiles of the normal range have a significantly increased risk of developing a number of the most common cancers, such as colon, breast, prostate and possibly lung. Indeed, it has been suggested that serum IGF-1 levels have a stronger association than most other risk factors for these cancers.[36] However, not all studies have shown this association, with one study reporting increased levels of IGFBP-3 in patients with colorectal cancer.[37] Certainly, causality has not yet been established and elevated serum IGF-1 levels in patients with cancer may simply be a surrogate measure of some other process. In order to clarify this issue, Renehan et al. recently performed a systematic review and meta-regression analysis of epidemiological case-control studies, including studies nested in cohorts, to investigate the association between concentrations of IGF-1 and IGFBP-3 and prostate, colorectal, premenopausal and postmenopausal breast and lung cancer.[38] The studies included in this analysis had to be published as full article findings expressed as odds ratios with 95% CI and reported an association for at least three categories of peptide concentration. Methodological quality was assessed by the use of published criteria for observational studies. Of 139 relevant publications identified, 21 (26 data sets) had outcome data appropriate for meta-analysis that included 3609 cases and 7137 controls.

High normal concentrations of IGF-1 were associated with a twofold increased risk of prostate cancer (P = 0·009), colorectal cancer (P = 0·09) and premenopausal breast cancer (P = 0·007) but not postmenopausal breast cancer or lung cancer. IGFBP-3 was associated positively with an increased risk of premenopausal breast cancer (P = 0·05). Mutual adjustment for IGF-1 and IGFBP-3 did not appear to affect results. The data confirm an undisputed link between IGF-1 and cancer. Whereas it is appreciated that IGFBP-3 influences IGF-1 bioactivity both by endocrine and by paracrine actions, and it also exerts IGF-independent effects on cell proliferation and apoptosis,[39,40] the link between IGFBP-3 levels and cancer risk is less clear cut.